Professional Documents
Culture Documents
Unit 5
Unit 5
Structure
5.0 Objectives
1 5.1 Introduction
5.2 Classical Valuation of Insurance Contracts
5.2.1 Life Insurance
5.2.2 Non-life lnsurance
5.0. OBJECTIVES
After going through this unit, you will be able to:
see the interplay between finance and insurance in risk valuation; and
Remember that the theory of insurance was formulated mainly to compute the
premiums for life insurance contracts. It was suggested that the methods of
evaluation of life insurance contracts could be carried on combining interest
and mortality. Daniel Bernoulli pointed out that risks should not be measured
by their expectations. He made use of gambling to argue that the preferences of
an individual might depend on her economic situation. Specifically, it could be
quite reasonable for a poorer individual to prefer an uncertain future payment
to another more uncertain payment with a large expected value. On the other
hand, a wealthier person would prefer the payment with the largest expected
value. Recall that we have already seen this argument while studying the
expected utility theory in MEC-01. We have also studied there that individuals
could to buy insurance contracts at a price, which might exceed the expected
value of the payments from the contract.
In order to draw insight from the similar decision making scenario, we may
look at the stock markets. Prices there undergo frequent fludtuations. Such a
feature can be described as a process akin to Brownian motion.
Samuelson (1965) developed a framework where the stock price was modelled
by a geometric Brownian motion i.e., the exponential function of a Brownian
motion. It had the advantage of not generating negative prices. Black and
Scholes (1 973) and Merton (1973) extended this framework with the additional
assumption that money could be deposited on a savings account. Their results
show that options on stocks should be priced such that no sure profits could
arise from composing portfolios of long and short positions in the underlying
stock and in the option itself. Cox, Ross and Rubinstein (1979) showed that the
change in the value of the stock between two trading times could attain two
different values. In that setting, they derived option prices and obtained the
pricing formulae of Black, Scholes and Merton as limiting cases by letting the
length of the time intervals between trading times tend to 0. Building on
concepts and ideas in Harrison and Kreps (1979) for discrete time models,
Harrison and Pliska (1981) gave a mathematical theory for pricing of options
under continuous trading and clarified the role of martingale theory in the
pricing of options and its connection to key concepts such as absence of
arbitrage h d completeness.
The net result of insights thrown by pricing in insurances and finance market in
the presence of risk is the schemes like catastrophe futures and options and
financial stop-loss reinsurance contracts. Thus. the recent methodology of
pricing principles in many insurance products in based on the theories behind
risk and combined approach of insurance and finance.
Valuing Risk
As you will see we have followed Merller (2001) in developing the present Management
unit. For an easy comprehension same notation as well as sequence of the
above work is retained. In the process we have borrowed freely from it.
Keeping in view the explorative nature of the theme required by the present
course we have discussed the basic ideas contained in Msller (2001). However,
for reference it will be useful to go through the entire work.
To appreciate the tools used for analysing the insurance valuation you should
have understood the stochastic calculus and financial mathematics we have
given in units 3 and 4.
A pure endowment contract with sum insured K and term T stipulates that
the amount K , the insurance benefit, is payable at time T contingent on
survival of the policy-holder. Assume that the contract is paid by a single
premium, say, at time 0. Assume further that the seiler of the contract (the
insurance company) invests the premium in some asset, which pays aerate of
return r = ( ) during [o,T] . For the ifh policyholder, the obligation of the
insurance company is given by the present value
J.ld:
H, = llTDrlKeU (2-1)
in case of the single premium. This principle is justified due to the law of large
numbers. As the size of the portfolio n is increased, the relative number of 1
n
x
1 "
survivors - = l{,; converges almost surely ( a x ) towards the probability
1-1.
,.
expected number, n p, .
Valuing Risk
net premium principle or the principle of equivalence A ( H ) = 0 ; Management
In addition to these, there are some other principles, which we list below:
That is, the expected utility of the final wealth V, + C ( H )- H from selling the
claim H at thk premium G ( H ) should equal the expected utility of Vo Here
Vo can be interpreted as the wealth associated with not selling the claim H and
th~refore,selling the claim -leaves the expected utility unaffected, i.e., it leads
nd~therto an increase nor a decrease in expectedzltiliEy.
*-
Exponential Principle
In practice use is made of the exponential principle, which is,obtained for the
exponential utility function
In particular, when Vo is constant P -a.s., the solution to (2.5) does not depend
on V, and is given by
c(H) = p E [ ~ ] + ( l - p ) ~ - ' ( l - ~ ) ,
.........................................................................................
at t and t7, ,is the discounted amount invested in the savings account. The'
discounted value at time t of pt is given by ( p )= 9, + q, . The strategy p is .
self-financing if
Let us consider the terms in (3.1). V0( p ) is the amount invested at time 0 and
[S,dX,is thp accumulated trading gains generated up to time t . Thus, for a
self-financing strategy p, the current value of the portfolio p, at time t is
exactly the initially invested amount plus trading gains, so that no in- or out-
flow of capital which has taken place during [ 0 , t ] .Let- H be interpreted as
payoff for some financial contract and investor such as an investment bank
who considers selling a contingent claim, i.e., Rr measurable random
Actuarial Techniques 1 variable H,. Then a contract (or claim) specifying the discounted payoff H at
time T is said to be attainable if there exists a self-financing strategy p such
that V, (p) = H a.s., that is, if H coincides with the terminal value of a self-
financing strategy. Thus, a claim is attainable if and oniy if it can be
represented as a constant H, plus a stochastic integral for the discounted
stock price process
V, = vo(a)+ j $.a,
- JW, .
The net result of all these steps is to find the estimation of gain E from the
trade. If the price of H is Ho + E, with E < 0 , the gain E can be obtained-by
buying H and using the hedging strategy p . We have shswh how the amount
H, received at time 0 can be transformed into the amount H at time T bj.
following a self-ftnancing strategy. There is -np further payment and hence no
more risk. This implies, H, is the fair price.
. ,
When the price v,* is charged and the strategy @ is adopted, the hedger can
generate an amount, which exceeds the needed amount H , P -a.s.
Consequently a major advantage of this approach is to have no risk to the
hedger. That is, there is no need of additional capital to pay the amount H to
the buyer of the contract after making the initial investment.
where the supremum is taken over all strategies 9 from some suitable space of
processes. The number W ( z ,p, c ) is the maximum obtainable expected utility
for an investor with initial capital cwhen she makes an investment in '
z 1 p units of the risk H . The fair price i2 ( H ;c ) of H is then defined as the
solution 3 to the equation
called the approximation price for H , and the optimal strategy is the mean-
variance hedging strategy.
Quantile hedging
.........................................................
L......... .......................
Let S, be the value of the stock index at time t . Following the notations used
in M ~ l l e r(2001), we refer to the entire development of the stock as S .
Consider a portfolio consisting of n policyholders with remaining life times
q...q. To simplify the analysis assume that they all buy the same form of unit-
linked pure endowment contract at time 0. Let the life times be stochastically
independent of the development on the financial market. The contracts specify
the payment of some (non-negative) amount f (s)to the policyholder at time
T if she is still alive at this time; the function f describes nature of the
dependence on the development of the stock price. Under such a set up, the
present value at time 0 of the insurance cohpany's liability towards the n
policyholders is given as I
From (4.2) and (4.3) above we get the contracts which are called pure unit-
linked contracts and (4.3) is called unit-linked with guarantee, the guaranteed
benefit being K . We may extend f to be a more complex function of the
'
process S . For example, we may have a guaranteed annual return
In such a case s,-sj-1 gives the return in year j on the asset S and S, is
s,-I
the guaranteed return in year j . The amount payable at time T is guaranteed at
the time of taking the contract (i.e., at time 0 ) against falling short of
K . n : ( 1 + 8 , ) . However, the guarantee goes beyond this in worst-case
scenario.
Valuing Risk
'Thus we see that the modern theories of finance offer new valuation principles
Management
and investment strategies for unit-linked insuranse contracts. It combines the
traditional (law of large numbers) arguments from life insurance with the
methods of Black and Scholes (1973) and Merton (1973). In the process we
could (i) replace the uncertainty of the insured lives by their expected values,
(ii) include the actual insurance claims by taking into account mortality risk as
well as financial risk. Such modified claims have contained financial
uncertainty only. Thus, instead of considering the claim (4.1), this method
examines
,
where the notation p,= P(T, > T) . Such modified claims can be treated as
options, with a long maturity period. As a result, these could be priced and
hedged using the basic principles of Black and Scholes (1973) and Merton
(1 973).
Let us look at the insurance related results of unit-link4 contracts:
In case of pure unit-linked contract (4.2), the claim (4.4) is proportional to the
terminal value of the stock ST and hence can be hedged by a buy-and-hold
strategy which consists in buying n,py units of the stock at time 0 and holding
these until time T .
Case of no guarantee
When there is no guarantee, the unique no-arbitrage price of H' is simply
!
I n,p,s,. Thus, you can prescribe one possible fair premium for each
I policyholder is ,-pysOwith the probability of survival to T times the value at
time 0 of the stock index.
Case of contract with benefit
i Write f (s)= max(S,, K ) = (S, - K)' +K and consider the pricing of a
Ii
European call option. Then from (4.4) it follows that the premium is r + p y ~ f ,
1 where v,' is the price at time 0 of the purely financial contract which pays
f ( S ) at time T
We have seen that CAT futures and Catastrophe-linked bonds are aimed at a
larger group of investors. The reinsurance contracts in these combine elements
of insurance and financial derivatives. There are some new contracts described
by 'Integrated Risk Management Solutions'. One example init is financial
stop-loss contract. It pays at some fixed time T the amount
where U,. is the aggregate 'claim amount during [0, TI on some insurance
portfolio, YT is some financial loss and K is some retention limit. The
financial stop-loss contracts cover large losses due to fluctuations within the
insurance portfolio (insurance risk) and adverse development of the financial
markets (financial risk). Reinsurance companies would be able to sell spreads
on the form
where 0 I K, IK, ,which covers the (K,, K,] layer of the losses U , + YT.
The insurance contract (4.5) helps provide cover for the insurer's total risk, i.e.,
the oornbined insurance risk from the insurance portfolio and the financial risk
from the financial portfolio. With a traditional stop-loss contract, the reinsurer
would cover insurance losses exceeding the level K . However, the financial
stop-loss contract is designed so that the cover is only psid provided that the
insurance loss augmented by the financial loss exceeds this level. Thus, a large
financial gain - YT may compensate for large insurance losses, and in this
situation, the buyer does not really need additional compensation from the
reinsurer.
Valuing Risk
5.4.3 Combining Theories for Financial and Actuarial Management
Valuation
Consider the basic difference between financial valuation techniques, (more
precisely, pricing by no-arbitrage) and the classical actuarial valuation
principles reviewed above. We have seen that the financial valuation
are formulated within a framework which includes the possibility of trading
certain assets. In contrast to these the classical actuarial valuation principles
are based on considerations involving the law of large numbers. In addition,
the financial valuation principles are based on dynamic trading, whereas many
decision problems in insurance are traditionally analysed taking a static view.
Gerber and Shiu (1996) among others have examined the situation where the
logarithm of the stock price process is a Levy process, i.e., a process with
independent and stationary increments. Within this setting, they demonstrate
how the Esscher transform (2.6) can be used in the pricing of options. They
also give a very simple option pricing formula which involves Esscher
transforms and which, for a European call option, specialises to the Black-
Scholes formula in the case of a geometric Brownian motion. Moreover, they
also demonstrate how this pricing formula can be derived through a simple
1-a
x
utility indifference argument in the case of power utility function u ( x ) = -
1-a
with parameter a > 0 .
In Schweizer (2001), the starting points are the traditional standard deviation
and variance principles, which are of the form (2.4). These principles are taken
as measures of riskiness, which assign to each claim a premium. It is then
argued that the measures can equivalently be viewed as measures of
preferences which operate on the insurer's terminal wealth by simply changing
the sign on the loading factor.
I
4
Complete market : A market is complete with respect to a trading strategy if
there exists a self-financing trading strategy such that at any time t, the returns
r of the two strategies are equal. That is, all cash flows for a trading strategy can
t be replicated by a similar synthetic trading strategy. Because a trading strategy
can be simplified into a set of simple contingent claims ( a trading strategy that
pays 1 in one state and zero in every other state ), a complete market can be
generalized as the ability to replicate c ~ flows h of all simple contingent
claims.
Hazard Function : By calculating the failure rate for smaller and smaller
intervals of time At, the interval becomes infinitesimally small. This results in
the hazard function, which is the instantaneous failure rate at any point in
time:
~(t)-~(t+~t)
h(t) = lim
+.o At.R(t)
~(tst)=~(t)=i-~(t),t>o.
~ ( t=) If ( x ) m .
Embrechts, Paul, Riidige Freg, Hansis & Furren (1999), Stochastic Process in
Insurance and Finance, ETH ziirich, Switzerland (see internet)